INDEX

Explanations

let's not/just/be

This neuron responds to conversational imperative lead-ins, especially “let’s” (and its variants like “let us”) plus associated words (e.g., “not,” “just,” “talk,” “say”), marking proposals or exhortations.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

we

-2.05

it

-1.95

 they

-1.87

you

-1.87

was

-1.74

が見えて

-1.74

 Another

-1.64

 Recently

-1.63

まわり

-1.59

 using

-1.54

POSITIVE LOGITS

zości

1.70

ｘ

1.60

 επίσης

1.55

푀

1.47

 Руси

1.46

 oxígeno

1.46

 сдела

1.44

尔

1.41

苹果

1.40

 även

1.39

Activations Density 0.004%